Longitudinal Image Registration Algorithm For Infrared Images For Chemotherapy Response Monitoring And Early Detection Of Breast Cancers

Abstract

This algorithm provides a marker-free approach to establishing the pixel correspondence among the IR images taken at different times, which is the basis for quantitatively characterizing the variation of the heat energy and patterns pixel-wise on a breast surface. The idea is to use the corner points of the heat pattern and the branch points of the skeletons of the heat pattern on the body surface as the initial fiducial points for the longitudinal IR image registration. The Thin-Plate Spline technique is used to model the nonlinear deformation between two IR images taken at two different times. Mutual information between the TPS-transformed image and the target image is employed as the metric quantifying the quality of the longitudinal IR image registration. To optimize the registration, Nelder-Mead simplex method is used to locally modify the pairings of the fiducial points in the source and target IR images to maximize the mutual information.

Description

FIELD OF THE INVENTION[0001]The invention relates to a method of image registration for assessing chemotherapy treatment response and early detection. More specifically, this invention relates to an image registration of longitudinal infrared images accepting MIR (middle-wave Infra-red) and LIR (long-wave Infra-red) over several time points in order to obtain more usable information.BACKGROUND OF THE INVENTION[0002]The breast cancer has been one of major cancer diseases leading death for a long time. It can effectively reduce menace of breast cancer by detecting breast cancer early and monitoring the effect of chemotherapy completely. However, there are not any methods achieving the goals effectively. Therefore, it is an important issue to detect the breast cancer early for the health of women.[0003]Images generate by medical modalities such as X-ray mammograms, MRI images and Ultrasound has the characteristic of showing distinct anatomic features such as nipples and boundaries. The...

AI Technical Summary

Benefits of technology

[0007]In the present application, the goal of image registration is to deform the difference geometry shapes of longitudinal infrared images accepting LIR and MIR alignment into the same structure and obtain more usable information. The image registration algorithm is separated into three steps: (1) Detecting feature points, especially, Harris corner detector algorithm is used to find corner points on infrared images and cross-points of the middle lines in the vessel heat patterns are found via Hessian matrix; (2) Searching corresponding points, we established the relationship of corner points and cross-points between the source image and the target image manually or automatically and form a set of corresponding point for Thin Plane Spline (TPS) model, where the target image is measured at the first time point as a reference image and the source image is measure after the target image for registering based on the target image; and (3) Dynamical correcting the locations of corresponding points, Nelder-Mead simplex method used to modify the locations of control points on the source image and maximize mutual information (MI). In this way, optimum and subpixel registration could be achieved by this step.

Problems solved by technology

However, there are not any methods achieving the goals effectively.

Therefore, it is an important issue to detect the breast cancer early for the health of women.

In contrast, infrared images have less distinct boundaries, therefore it is challenging to detect feature points for image registration and making infrared images alignment a difficult hurdle to cross.

Nevertheless, the usefulness of IR images in detecting breast cancers at a single time is not usually enough to assess the chemotherapy treatment response and early detection due to the physiological and environmental influence on the skin temperature distribution.

Images